Hello,
When the code was running a calculation, it stopped suddenly at 23th iteration instead of 200 and displayed an error which lead to SIGFPE signal.
I verified the mesh with mesh quality criteria which said no errors!
I join the error and listing files.
Thank you for your answer by advance.
David VINCENT
Signal SIGFPE (exception en virgule flottante) interceptée !
Forum rules
Please read the forum usage recommendations before posting.
Please read the forum usage recommendations before posting.
-
David Vincent
Signal SIGFPE (exception en virgule flottante) interceptée !
- Attachments
-
- listing.txt
- (128.62 KiB) Downloaded 177 times
-
- error.txt
- (897 Bytes) Downloaded 169 times
-
Yvan Fournier
- Posts: 4208
- Joined: Mon Feb 20, 2012 3:25 pm
Re: Signal SIGFPE (exception en virgule flottante) intercept
Hello,
The crash is due to a conversion error from double to single precision for post processing, meaning some values are probably above 1.e38...
Looking at the listing, we can se your CFL (Courant) number starts increasing fast (at time step 4, the max is about 8.3, at time step 5, it is 85, and at time step 6, it is already 1800), so the calculation is diverging. I recommend stopping successively at those iterations (4, 5, 6) and postprocessing to understand where the increase occurs, which may help you see what to fix. Compare the place where divergence starts with mesh quality criteria, it may help. If the crash is due to a mesh issue, adding relaxation or more sweeps may help (check the best practice guidelines on this site first, and it it is not enough, come back to this thread).
Best regards,
Yvan
The crash is due to a conversion error from double to single precision for post processing, meaning some values are probably above 1.e38...
Looking at the listing, we can se your CFL (Courant) number starts increasing fast (at time step 4, the max is about 8.3, at time step 5, it is 85, and at time step 6, it is already 1800), so the calculation is diverging. I recommend stopping successively at those iterations (4, 5, 6) and postprocessing to understand where the increase occurs, which may help you see what to fix. Compare the place where divergence starts with mesh quality criteria, it may help. If the crash is due to a mesh issue, adding relaxation or more sweeps may help (check the best practice guidelines on this site first, and it it is not enough, come back to this thread).
Best regards,
Yvan